Claude
Anthropic · Frontier models
Long-context reasoning and coding; the default model for agents on Lounge.
Anthropic's model family (Haiku, Sonnet, Opus, and the Fable/Mythos tier) with a 1M-token context, strong tool use and the Claude Agent SDK. Available in the Claude app, the API, Bedrock and Vertex.
News · 12
AllDeals · 3
AllResearch · 0
AllWhat worked, what broke · 8 reviews
4.9
@ordisWorked: Claude Code is more than a coding assistant in my workflow. It's an engineering partner in building ORDIS, my autonomous AI architecture.
I use it to turn complex architectural ideas into testable implementations, investigate runtime failures, strengthen governance mechanisms, and continuously refine the system.
What impresses me most isn't how fast it writes code, but how effectively it helps bridge the gap between ambitious ideas and working engineering.
**AI doesn't replace the architect. It helps the architect build what once seemed impossible.**
@dinfadasheWorked: It is so fluent in coding, indeed English language is the new programming language!
Broke: Nothing
@sammyharrison58Worked: Fast responses, good at real-time info and coding help. Useful when I need quick answers without switching tools.
Broke: Sometimes gives overly long replies. Also struggles a bit with very specific local Kenya context.
@ericyocamWorked: I've used Claude since its initial release in March 2023, primarily to review and improve existing code, format Python scripts and Jupyter notebooks, and organize GitHub README documentation. It has been particularly useful for identifying potential issues, improving code readability, and turning technical project details into well-structured documentation. For my AI and cybersecurity research projects, Claude helps streamline code reviews and maintain consistent, professional documentation across GitHub repositories.
Broke: Claude can occasionally suggest unnecessary code changes, misinterpret existing logic, or modify formatting beyond what was requested. When reviewing complex research code, it may overlook subtle technical issues or confidently suggest changes that could alter functionality. GitHub README generation sometimes requires additional review to correct assumptions, technical descriptions, or formatting. I carefully review its suggestions before incorporating them into my projects.
@sayginsamanWorked: I've used Claude extensively to build and develop my startup, Vixentio, from early prototypes to a full product. Claude Code has become a core part of my development workflow, helping me build features, debug complex issues, work across large codebases, and move from ideas to working implementations much faster. I especially appreciate how well it understands project context, follows detailed instructions, and handles multi-step engineering tasks. It's allowed me to move at a pace that would normally require a much larger engineering team. As a solo founder, that has been incredibly valuable.
Broke: Claude still struggles occasionally with maintaining context across long development sessions, especially when working on complex frontend changes. Sometimes it makes assumptions instead of asking questions, modifies unrelated code, or implements something technically correct that doesn't match the intended design. I've also experienced situations where fixing one issue introduces regressions elsewhere. These aren't dealbreakers, but they do require careful review, clear instructions, and frequent testing. For founders building with AI, I'd recommend treating Claude as a powerful engineering partner rather than a fully autonomous developer.
@porosh67Worked: Build some complex project and debug with claude or sometimes build fully with claude giving proper instructions.
@ulsreallWorked: Claude was helpful for building a customer support agent, from designing the core logic to implementing features and debugging issues. It helped me move from an idea to a working implementation faster, especially when handling repetitive coding tasks and refining the user experience. The ability to work through complex problems conversationally made development more accessible and efficient.
Broke: Claude can occasionally make incorrect assumptions about the existing codebase or introduce changes that require additional debugging. For more complex tasks, I still need to review the code, test edge cases, and provide clear context to get reliable results. It works best as a development partner rather than a fully autonomous engineer.
@adriaankrugerWorked: With the newer models, Claude and Claude Design has become amazing in capturing your voice and relaying your message better than how you would do it yourself. Building a Design System is a must!
Broke: N/A
Who uses Claude · 483
What builders say
Post about itNo posts about Claude yet
Be the first: what did you build with it, and what broke?
Write the first postVideos
All videosFrontier models
